Objective: As we all know, there would be some distortions of electrodes or even distortions of patient’s brain images when matching the brain atlases with preoperative images before Deep Brain Stimulation (DBS) surgical treatment in previous studies [1,2,3]. In our research, we normalized these clinical images to the coordinate space of the atlas, followed by coarse registration and accurate registration. Finally, the high precision matching between clinical data and brain atlas data was achieved.

Background: DBS surgery is an important treatment for the patients with Parkinson’s disease. Before DBS surgery, the surgical planning and preoperative navigation are needed. Preoperative navigation is a process of identifying deep brain tissues, also is a key part that directly determines the effect of the surgery. Human brain atlas is a powerful tool to reflect the anatomical location and shape information of human brain tissue. Thus, the registration between brain atlases and preoperative images is an effective method to realize accurate preoperative navigation and nucleus recognition before DBS treatment.

Method: In this research, we chose ICBM-152 atlas, which was widely used and had many sub-atlases, as the standard atlas data. Based on this, we studied the registration of preoperative images and brain atlases. First of all, the preoperative images were standardized to the coordinate space of the atlas, and then the piecewise linear registration was used to coarsely map the spectrum, so that the sizes, the shapes and the outer contours of the two could be in good agreement. At last, non-rigid registration of the brain atlas with preoperative data was achieved using mutual information as a measure of similarity.

Results: This method can calculate personalized brain atlas for each patient. The experimental results show that the mutual information between the original clinical image and the brain atlas can be increased to 1.217 compared with 0.558 in the initial stage of the experiment.

Conclusion: In this paper, we made mid-sagittal correction of preoperative images, and this enabled the images and ICBM-152 atlases consistent in space. After that, we made a further correction of the images with the method named of the piecewise linear registration, and followed with non-rigid registration. By doing those, we successfully realized the accurate matching between the preoperative images and ICBM-152 atlases.

References: [1] Andreas H,Kühn AA.Lead-DBS: a toolbox for deep brain stimulation electrode localizations and visualizations[J]. Neuroimage, 2015, 107(107):127-135. [2] Videen TO,Campbell MC,Tabbal SD,et al.Validation of a fiducial-based atlas localization method for deep brain stimulation contacts in the area of the subthalamic nucleus[J].J Neurosci Meth,2008,168(2):275. [3] Silva NMD,Rozanski VE,Cunha JPS.A 3D multimodal approach to precisely locate DBS electrodes in the basal ganglia brain region[A].International IEEE/EMBS Conference on Neural Engineering, IEEE[C].2015.

« Back to 2019 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/registration-algorithm-of-dbs-preoperative-image-and-icbm-152-atlas/